Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
0.0.19
af8ff9a2
·
add ability to skip prompt asking whether to clear previous results, by...
·
Feb 10, 2023
0.0.18
df17dcd0
·
need torch 1.12 for complex stft discriminator from audiolm
·
Feb 09, 2023
0.0.17
f2dc4628
·
in contrastive learning, make sure always the full batch
·
Feb 08, 2023
0.0.16
0d1f2b3d
·
batch size must be greater than 1 for contrastive learning
·
Feb 08, 2023
0.0.15
6d45461a
·
helpful assert
·
Feb 07, 2023
0.0.14
846740c3
·
forget about weight decay, and make sure to clip wavs by data_max_length
·
Feb 07, 2023
0.0.12
bb73ba7a
·
curtail to shortest, as no self attention masking strategy in audio spectrogram transformer yet
·
Feb 06, 2023
0.0.11
c138c935
·
get a mulan trainer out, remove wip
·
Feb 06, 2023
0.0.10
7dcfc974
·
add patch dropout
·
Feb 05, 2023
0.0.9
5aaee642
·
adopt dual patchnorm
·
Feb 03, 2023
0.0.8
26c82682
·
ok it is working
·
Feb 02, 2023
0.0.7
e96259b7
·
it is done
·
Feb 02, 2023
0.0.6
6a9e4262
·
use base class from audiolm-pytorch for type validation
·
Feb 02, 2023
0.0.5
0503f7d1
·
override parameters, so if mulan embed quantizer is passed to audiolm...
·
Feb 02, 2023
0.0.4
c40aee30
·
working inference example
·
Feb 02, 2023
0.0.3
5a950d50
·
complete the residual vector quantization of the joint embedding music-text...
·
Feb 01, 2023
0.0.2
ff4c75f8
·
add decoupled contrastive learning for mulan as an option
·
Jan 31, 2023
0.0.1
9a0554f4
·
complete mulan, lay out a gameplan for remaining week
·
Jan 31, 2023
Prev
1
2
Next